A Pipeline for the Implementation and Visualization of Explainable Machine Learning for Medical Imaging Using Radiomics Features

Current state and promise of user-centered design to harness explainable AI in clinical decision-support systems for patients with CNS tumors

In neuro-oncology, MR imaging is crucial for obtaining detailed brain images to identify neoplasms, plan treatment, guide surgical intervention, and monitor the tumor's response. Recent AI advances in neuroimaging have promising applications in neuro-oncology, including guiding clinical decisions and improving patient management. However, the lack of clarity on how AI arrives at predictions has hindered its clinical translation. Explainable AI (XAI) methods aim to improve trustworthiness and informativeness, but their success depends on considering end-users' (clinicians') specific context and preferences. User-Centered Design (UCD) prioritizes user needs in an iterative design process, involving users throughout, providing an opportunity to design XAI systems tailored to clinical neuro-oncology. This review focuses on the intersection of MR imaging interpretation for neuro-oncology patient management, explainable AI for clinical decision support, and user-centered design. We provide a resource that organizes the necessary concepts, including design and evaluation, clinical translation, user experience and efficiency enhancement, and AI for improved clinical outcomes in neuro-oncology patient management. We discuss the importance of multi-disciplinary skills and user-centered design in creating successful neuro-oncology AI systems. We also discuss how explainable AI tools, embedded in a human-centered decision-making process and different from fully automated solutions, can potentially enhance clinician performance. Following UCD principles to build trust, minimize errors and bias, and create adaptable software has the promise of meeting the needs and expectations of healthcare professionals.

Development of a flexible feature selection framework in radiomics-based prediction modeling: Assessment with four real-world datasets

There are several important challenges in radiomics research; one of them is feature selection. Since many quantitative features are non-informative, feature selection becomes essential. Feature selection methods have been mixed with filter, wrapper, and embedded methods without a rule of thumb. This study aims to develop a framework for optimal feature selection in radiomics research. We developed the framework that the optimal features were selected to quickly through controlling relevance and redundancy among features. A 'FeatureMap' was generated containing information for each step and used as a platform. Through this framework, we can explore the optimal combination of radiomics features and evaluate the predictive performance using only selected features. We assessed the framework using four real datasets. The FeatureMap generated 6 combinations, with the number of features selected varying for each combination. The predictive models obtained high performances; the highest test area under the curves (AUCs) were 0.792, 0.820, 0.846 and 0.738 in the cross-validation method, respectively. We developed a flexible framework for feature selection methods in radiomics research and assessed its usefulness using various real-world data. Our framework can assist clinicians in efficiently developing predictive models based on radiomics.

radMLBench: A dataset collection for benchmarking in radiomics

BACKGROUND

New machine learning methods and techniques are frequently introduced in radiomics, but they are often tested on a single dataset, which makes it challenging to assess their true benefit. Currently, there is a lack of a larger, publicly accessible dataset collection on which such assessments could be performed. In this study, a collection of radiomics datasets with binary outcomes in tabular form was curated to allow benchmarking of machine learning methods and techniques.

METHODS

A variety of journals and online sources were searched to identify tabular radiomics data with binary outcomes, which were then compiled into a homogeneous data collection that is easily accessible via Python. To illustrate the utility of the dataset collection, it was applied to investigate whether feature decorrelation prior to feature selection could improve predictive performance in a radiomics pipeline.

RESULTS

A total of 50 radiomic datasets were collected, with sample sizes ranging from 51 to 969 and 101 to 11165 features. Using this data, it was observed that decorrelating features did not yield any significant improvement on average.

CONCLUSIONS

A large collection of datasets, easily accessible via Python, suitable for benchmarking and evaluating new machine learning techniques and methods was curated. Its utility was exemplified by demonstrating that feature decorrelation prior to feature selection does not, on average, lead to significant performance gains and could be omitted, thereby increasing the robustness and reliability of the radiomics pipeline.

Towards Improved XAI-Based Epidemiological Research into the Next Potential Pandemic

By applying AI techniques to a variety of pandemic-relevant data, artificial intelligence (AI) has substantially supported the control of the spread of the SARS-CoV-2 virus. Along with this, epidemiological machine learning studies of SARS-CoV-2 have been frequently published. While these models can be perceived as precise and policy-relevant to guide governments towards optimal containment policies, their black box nature can hamper building trust and relying confidently on the prescriptions proposed. This paper focuses on interpretable AI-based epidemiological models in the context of the recent SARS-CoV-2 pandemic. We systematically review existing studies, which jointly incorporate AI, SARS-CoV-2 epidemiology, and explainable AI approaches (XAI). First, we propose a conceptual framework by synthesizing the main methodological features of the existing AI pipelines of SARS-CoV-2. Upon the proposed conceptual framework and by analyzing the selected epidemiological studies, we reflect on current research gaps in epidemiological AI toolboxes and how to fill these gaps to generate enhanced policy support in the next potential pandemic.

Model-agnostic explanations for survival prediction models

Advanced machine learning methods capable of capturing complex and nonlinear relationships can be used in biomedical research to accurately predict time-to-event outcomes. However, these methods have been criticized as "black boxes" that are not interpretable and thus are difficult to trust in making important clinical decisions. Explainable machine learning proposes the use of model-agnostic explainers that can be applied to predictions from any complex model. These explainers describe how a patient's characteristics are contributing to their prediction, and thus provide insight into how the model is arriving at that prediction. The specific application of these explainers to survival prediction models can be used to obtain explanations for (i) survival predictions at particular follow-up times, and (ii) a patient's overall predicted survival curve. Here, we present a model-agnostic approach for obtaining these explanations from any survival prediction model. We extend the local interpretable model-agnostic explainer framework for classification outcomes to survival prediction models. Using simulated data, we assess the performance of the proposed approaches under various settings. We illustrate application of the new methodology using prostate cancer data.

From pixels to prognosis: unveiling radiomics models with SHAP and LIME for enhanced interpretability

Radiomics-based prediction models have shown promise in predicting Radiation Pneumonitis (RP), a common adverse outcome of chest irradiation. Τhis study looks into more than just RP: it also investigates a bigger shift in the way radiomics-based models work. By integrating multi-modal radiomic data, which includes a wide range of variables collected from medical images including cutting-edge PET/CT imaging, we have developed predictive models that capture the intricate nature of illness progression. Radiomic features were extracted using PyRadiomics, encompassing intensity, texture, and shape measures. The high-dimensional dataset formed the basis for our predictive models, primarily Gradient Boosting Machines (GBM)-XGBoost, LightGBM, and CatBoost. Performance evaluation metrics, including Multi-Modal AUC-ROC, Sensitivity, Specificity, and F1-Score, underscore the superiority of the Deep Neural Network (DNN) model. The DNN achieved a remarkable Multi-Modal AUC-ROC of 0.90, indicating superior discriminatory power. Sensitivity and specificity values of 0.85 and 0.91, respectively, highlight its effectiveness in detecting positive occurrences while accurately identifying negatives. External validation datasets, comprising retrospective patient data and a heterogeneous patient population, validate the robustness and generalizability of our models. The focus of our study is the application of sophisticated model interpretability methods, namely SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-Agnostic Explanations), to improve the clarity and understanding of predictions. These methods allow clinicians to visualize the effects of features and provide localized explanations for every prediction, enhancing the comprehensibility of the model. This strengthens trust and collaboration between computational technologies and medical competence. The integration of data-driven analytics and medical domain expertise represents a significant shift in the profession, advancing us from analyzing pixel-level information to gaining valuable prognostic insights.

Explaining decisions of a light-weight deep neural network for real-time coronary artery disease classification in magnetic resonance imaging

In certain healthcare settings, such as emergency or critical care units, where quick and accurate real-time analysis and decision-making are required, the healthcare system can leverage the power of artificial intelligence (AI) models to support decision-making and prevent complications. This paper investigates the optimization of healthcare AI models based on time complexity, hyper-parameter tuning, and XAI for a classification task. The paper highlights the significance of a lightweight convolutional neural network (CNN) for analysing and classifying Magnetic Resonance Imaging (MRI) in real-time and is compared with CNN-RandomForest (CNN-RF). The role of hyper-parameter is also examined in finding optimal configurations that enhance the model's performance while efficiently utilizing the limited computational resources. Finally, the benefits of incorporating the XAI technique (e.g. GradCAM and Layer-wise Relevance Propagation) in providing transparency and interpretable explanations of AI model predictions, fostering trust, and error/bias detection are explored. Our inference time on a MacBook laptop for 323 test images of size 100x100 is only 2.6 sec, which is merely 8 milliseconds per image while providing comparable classification accuracy with the ensemble model of CNN-RF classifiers. Using the proposed model, clinicians/cardiologists can achieve accurate and reliable results while ensuring patients' safety and answering questions imposed by the General Data Protection Regulation (GDPR). The proposed investigative study will advance the understanding and acceptance of AI systems in connected healthcare settings.

A scoping review of interpretability and explainability concerning artificial intelligence methods in medical imaging

PURPOSE

To review eXplainable Artificial Intelligence/(XAI) methods available for medical imaging/(MI).

METHOD

A scoping review was conducted following the Joanna Briggs Institute's methodology. The search was performed on Pubmed, Embase, Cinhal, Web of Science, BioRxiv, MedRxiv, and Google Scholar. Studies published in French and English after 2017 were included. Keyword combinations and descriptors related to explainability, and MI modalities were employed. Two independent reviewers screened abstracts, titles and full text, resolving differences through discussion.

RESULTS

228 studies met the criteria. XAI publications are increasing, targeting MRI (n = 73), radiography (n = 47), CT (n = 46). Lung (n = 82) and brain (n = 74) pathologies, Covid-19 (n = 48), Alzheimer's disease (n = 25), brain tumors (n = 15) are the main pathologies explained. Explanations are presented visually (n = 186), numerically (n = 67), rule-based (n = 11), textually (n = 11), and example-based (n = 6). Commonly explained tasks include classification (n = 89), prediction (n = 47), diagnosis (n = 39), detection (n = 29), segmentation (n = 13), and image quality improvement (n = 6). The most frequently provided explanations were local (78.1 %), 5.7 % were global, and 16.2 % combined both local and global approaches. Post-hoc approaches were predominantly employed. The used terminology varied, sometimes indistinctively using explainable (n = 207), interpretable (n = 187), understandable (n = 112), transparent (n = 61), reliable (n = 31), and intelligible (n = 3).

CONCLUSION

The number of XAI publications in medical imaging is increasing, primarily focusing on applying XAI techniques to MRI, CT, and radiography for classifying and predicting lung and brain pathologies. Visual and numerical output formats are predominantly used. Terminology standardisation remains a challenge, as terms like "explainable" and "interpretable" are sometimes being used indistinctively. Future XAI development should consider user needs and perspectives.

Assessment of brain cancer atlas maps with multimodal imaging features

BACKGROUND

Glioblastoma Multiforme (GBM) is a fast-growing and highly aggressive brain tumor that invades the nearby brain tissue and presents secondary nodular lesions across the whole brain but generally does not spread to distant organs. Without treatment, GBM can result in death in about 6 months. The challenges are known to depend on multiple factors: brain localization, resistance to conventional therapy, disrupted tumor blood supply inhibiting effective drug delivery, complications from peritumoral edema, intracranial hypertension, seizures, and neurotoxicity.

MAIN TEXT

Imaging techniques are routinely used to obtain accurate detections of lesions that localize brain tumors. Especially magnetic resonance imaging (MRI) delivers multimodal images both before and after the administration of contrast, which results in displaying enhancement and describing physiological features as hemodynamic processes. This review considers one possible extension of the use of radiomics in GBM studies, one that recalibrates the analysis of targeted segmentations to the whole organ scale. After identifying critical areas of research, the focus is on illustrating the potential utility of an integrated approach with multimodal imaging, radiomic data processing and brain atlases as the main components. The templates associated with the outcome of straightforward analyses represent promising inference tools able to spatio-temporally inform on the GBM evolution while being generalizable also to other cancers.

CONCLUSIONS

The focus on novel inference strategies applicable to complex cancer systems and based on building radiomic models from multimodal imaging data can be well supported by machine learning and other computational tools potentially able to translate suitably processed information into more accurate patient stratifications and evaluations of treatment efficacy.

Explainable machine learning via intra-tumoral radiomics feature mapping for patient stratification in adjuvant chemotherapy for locoregionally advanced nasopharyngeal carcinoma

PURPOSE

This study aimed to discover intra-tumor heterogeneity signature and validate its predictive value for adjuvant chemotherapy (ACT) following concurrent chemoradiotherapy (CCRT) in locoregionally advanced nasopharyngeal carcinoma (LA-NPC).

MATERIALS AND METHODS

397 LA-NPC patients were retrospectively enrolled. Pre-treatment contrast-enhanced T1-weighted (CET1-w) MR images, clinical variables, and follow-up were retrospectively collected. We identified single predictive radiomic feature from primary gross tumor volume (GTVnp) and defined predicted subvolume by calculating voxel-wised feature mapping and within GTVnp. We independently validate predictive value of identified feature and associated predicted subvolume.

RESULTS

Only one radiomic feature, gldm_DependenceVariance in 3 mm-sigma LoG-filtered image, was discovered as a signature. In the high-risk group determined by the signature, patients received CCRT + ACT achieved 3-year disease free survival (DFS) rate of 90% versus 57% (HR, 0.20; 95%CI, 0.05-0.94; P = 0.007) for CCRT alone. The multivariate analysis showed patients receiving CCRT + ACT had a HR of 0.21 (95%CI: 0.06-0.68, P = 0.009) for DFS compared to those receiving CCRT alone. The predictive value can also be generalized to the subvolume with multivariate HR of 0.27 (P = 0.017) for DFS.

CONCLUSION

The signature with its heterogeneity mapping could be a reliable and explainable ACT decision-making tool in clinical practice.

Machine-Learning-Based Radiomic Features on Pre-Ablation MRI as Predictors of Pathologic Response in Patients with Hepatocellular Carcinoma Who Underwent Hepatic Transplant

Background: The aim was to investigate the role of pre-ablation tumor radiomics in predicting pathologic treatment response in patients with early-stage hepatocellular carcinoma (HCC) who underwent liver transplant. Methods: Using data collected from 2005–2015, we included adult patients who (1) had a contrast-enhanced MRI within 3 months prior to ablation therapy and (2) underwent liver transplantation. Demographics were obtained for each patient. The treated hepatic tumor volume was manually segmented on the arterial phase T1 MRI images. A vector with 112 radiomic features (shape, first-order, and texture) was extracted from each tumor. Feature selection was employed through minimum redundancy and maximum relevance using a training set. A random forest model was developed based on top radiomic and demographic features. Model performance was evaluated by ROC analysis. SHAP plots were constructed in order to visualize feature importance in model predictions. Results: Ninety-seven patients (117 tumors, 31 (32%) microwave ablation, 66 (68%) radiofrequency ablation) were included. The mean model for end-stage liver disease (MELD) score was 10.5 ± 3. The mean follow-up time was 336.2 ± 179 days. Complete response on pathology review was achieved in 62% of patients at the time of transplant. Incomplete pathologic response was associated with four features: two first-order and two GLRM features using univariate logistic regression analysis (p < 0.05). The random forest model included two radiomic features (diagnostics maximum and first-order maximum) and four clinical features (pre-procedure creatinine, pre-procedure albumin, age, and gender) achieving an AUC of 0.83, a sensitivity of 82%, a specificity of 67%, a PPV of 69%, and an NPV of 80%. Conclusions: Pre-ablation MRI radiomics could act as a valuable imaging biomarker for the prediction of tumor pathologic response in patients with HCC.

Adaptive Aquila Optimizer with Explainable Artificial Intelligence-Enabled Cancer Diagnosis on Medical Imaging

Explainable Artificial Intelligence (XAI) is a branch of AI that mainly focuses on developing systems that provide understandable and clear explanations for their decisions. In the context of cancer diagnoses on medical imaging, an XAI technology uses advanced image analysis methods like deep learning (DL) to make a diagnosis and analyze medical images, as well as provide a clear explanation for how it arrived at its diagnoses. This includes highlighting specific areas of the image that the system recognized as indicative of cancer while also providing data on the fundamental AI algorithm and decision-making process used. The objective of XAI is to provide patients and doctors with a better understanding of the system's decision-making process and to increase transparency and trust in the diagnosis method. Therefore, this study develops an Adaptive Aquila Optimizer with Explainable Artificial Intelligence Enabled Cancer Diagnosis (AAOXAI-CD) technique on Medical Imaging. The proposed AAOXAI-CD technique intends to accomplish the effectual colorectal and osteosarcoma cancer classification process. To achieve this, the AAOXAI-CD technique initially employs the Faster SqueezeNet model for feature vector generation. As well, the hyperparameter tuning of the Faster SqueezeNet model takes place with the use of the AAO algorithm. For cancer classification, the majority weighted voting ensemble model with three DL classifiers, namely recurrent neural network (RNN), gated recurrent unit (GRU), and bidirectional long short-term memory (BiLSTM). Furthermore, the AAOXAI-CD technique combines the XAI approach LIME for better understanding and explainability of the black-box method for accurate cancer detection. The simulation evaluation of the AAOXAI-CD methodology can be tested on medical cancer imaging databases, and the outcomes ensured the auspicious outcome of the AAOXAI-CD methodology than other current approaches.

Survey of Explainable AI Techniques in Healthcare

Artificial intelligence (AI) with deep learning models has been widely applied in numerous domains, including medical imaging and healthcare tasks. In the medical field, any judgment or decision is fraught with risk. A doctor will carefully judge whether a patient is sick before forming a reasonable explanation based on the patient's symptoms and/or an examination. Therefore, to be a viable and accepted tool, AI needs to mimic human judgment and interpretation skills. Specifically, explainable AI (XAI) aims to explain the information behind the black-box model of deep learning that reveals how the decisions are made. This paper provides a survey of the most recent XAI techniques used in healthcare and related medical imaging applications. We summarize and categorize the XAI types, and highlight the algorithms used to increase interpretability in medical imaging topics. In addition, we focus on the challenging XAI problems in medical applications and provide guidelines to develop better interpretations of deep learning models using XAI concepts in medical image and text analysis. Furthermore, this survey provides future directions to guide developers and researchers for future prospective investigations on clinical topics, particularly on applications with medical imaging.

Radiomics for Alzheimer's Disease: Fundamental Principles and Clinical Applications

Alzheimer's disease is a neurodegenerative disease with a huge impact on people's quality of life, life expectancy, and morbidity. The ongoing prevalence of the disease, in conjunction with an increased financial burden to healthcare services, necessitates the development of new technologies to be employed in this field. Hence, advanced computational methods have been developed to facilitate early and accurate diagnosis of the disease and improve all health outcomes. Artificial intelligence is now deeply involved in the fight against this disease, with many clinical applications in the field of medical imaging. Deep learning approaches have been tested for use in this domain, while radiomics, an emerging quantitative method, are already being evaluated to be used in various medical imaging modalities. This chapter aims to provide an insight into the fundamental principles behind radiomics, discuss the most common techniques alongside their strengths and weaknesses, and suggest ways forward for future research standardization and reproducibility.

Subtypes and Mechanisms of Hypertrophic Cardiomyopathy Proposed by Machine Learning Algorithms

Hypertrophic cardiomyopathy (HCM) is a relatively common inherited cardiac disease that results in left ventricular hypertrophy. Machine learning uses algorithms to study patterns in data and develop models able to make predictions. The aim of this study is to identify HCM subtypes and examine the mechanisms of HCM using machine learning algorithms. Clinical and laboratory findings of 143 adult patients with a confirmed diagnosis of nonobstructive HCM are analyzed; HCM subtypes are determined by clustering, while the presence of different HCM features is predicted in classification machine learning tasks. Four clusters are determined as the optimal number of clusters for this dataset. Models that can predict the presence of particular HCM features from other genotypic and phenotypic information are generated, and subsets of features sufficient to predict the presence of other features of HCM are determined. This research proposes four subtypes of HCM assessed by machine learning algorithms and based on the overall phenotypic expression of the participants of the study. The identified subsets of features sufficient to determine the presence of particular HCM aspects could provide deeper insights into the mechanisms of HCM.